design-systems

Warn

Audited by Gen Agent Trust Hub on Apr 24, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill documentation instructs the user to execute @southleft/design-systems-mcp using npx. This involves fetching and running code from an external package registry (NPM) from a source not identified as a pre-approved trusted vendor.
  • [COMMAND_EXECUTION]: The skill utilizes shell commands (npx) to initialize MCP servers. These commands execute third-party binaries that have the capability to perform arbitrary operations on the host system.
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it processes data and code generated from external APIs and design system repositories.
  • Ingestion points: Data entering the agent context via mcp__design_systems__search_components and mcp__magic__create_component (SKILL.md).
  • Boundary markers: The instructions do not specify any delimiters or warnings to ignore instructions embedded within the fetched component data or tokens.
  • Capability inventory: The skill is designed to generate React components and CSS (SKILL.md).
  • Sanitization: There is no evidence of sanitization or integrity verification for the code generated by external AI services or the metadata retrieved from design system libraries.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 24, 2026, 09:27 AM